Gonçalo Esteves / building capable systems Field Notes →

The Vault / People

Person

Palmer Luckey

founder of Oculus VR and of Anduril; guest at Moonshots Live (Moonshots 295 — Palmer Luckey: Autonomous Weapons Are Ancient and Why Anduril Won't Build Humanoids, recorded 2026-09-25, aired 2026-09-28, "Palmer Luckey: Autonomous Weapons Are Ancient and Why Anduril Won't Build Hum

status · researched added 8 Oct 2026 updated 18 Oct 2026 from Moonshots #295 · moonshots 169 · Moonshots #299 personorganizationspace see it in the graph →

What he is here for: founder of Oculus VR and of Anduril; guest at Moonshots Live (Moonshots #295, recorded 2026-09-25, aired 2026-09-28, "Palmer Luckey: Autonomous Weapons Are Ancient and Why Anduril Won't Build Humanoids"), and earlier the solo guest of moonshots 169 (2025-05-06). Named by the Pentagon on 2026-09-30 as a co-lead of Project Meridian (confirmed, see that page). Caption mangles: "Palmer Lucky", "Common Lucky", "Androl" / "Andoril" / "Andrew" (Anduril).

Why it is in this archive

Gate: guest of two full episodes (EP169, EP295) and cited in 21 held transcripts (12 backfill). Promoted 2026-10-08. EP295 decomposed 2026-10-18 (the research archive).

What he says on tape (EP295, timestamps from the .srt)

  • Autonomy has to live forward: if the central compute is jammed, intercepted or hacked "you're done"; under water "you can't talk to them… Having radio emissions actually exposes you" (0:08:13–0:09:31).
  • Autonomous weapons are old: Aegis "has a fully autonomous mode", CIWS and SeaRAM the same, Vietnam-era missiles that hunted radar emitters; then pits, poison traps, mines, torpedoes, golems: "we've had autonomous weapons for a very, very long time" (0:09:52–0:11:33). "That decision is going to remain with people" (0:10:34). The same lecture, same order, is on EP169 1:14:50–1:16:03.
  • Weapon models are not chat models: "they're largely non-deterministic. They're non-auditable… I need to process raw data, radar data feeds… sonar data from passive listening stations" (0:12:13–0:12:49). "We've been doing it since 2017" (0:12:00).
  • Specialisation over convergence in defense: "if I have a jet that has a reliable 5% every time edge over the enemy, that's worth paying a lot of money for" (0:15:34); robot submarines that "dive to 6,000 meter depth and… sit on the bottom of the ocean for months" (0:15:57).
  • What Anduril won't build: humanoids ("I've opted out because… I believe in them" 0:17:24), small quadcopter attack drones ("Too many other companies doing a great job" 0:18:27), space launch ("we operate a lot of our own satellites… but I don't want to be a launch provider" 0:18:34). Humanoids as the interface to legacy systems ("radar systems that cost… hundreds of millions… Nobody wants to man them during a hot conflict" 0:18:06). Consistent with EP169 0:59:16 and Dave's relay on Moonshots #283 1:46:31.
  • Procurement: "about 80% of… major defense acquisition programs… went to five companies" (0:20:34, ⚠️ public figure is >74% in 2023, secondary); Anduril is "less of a contractor and more of a product company. We use our own money to build products and then we sell them" (0:20:53). That route assumes private capital at a scale a small company does not have. Pitch deck page one: "Andoril will save taxpayers hundreds of billions of dollars a year by making tens of billions of dollars a year" (0:21:40).
  • P(doom): "I'm very, very worried about evil people using moderately competent AI to do evil things" (0:05:29); scorn for people "new to the dooming hobby" (0:05:36). Same view on EP169 ("It's not artificial intelligence. It's human stupidity I'm worried about" 1:23:08).
  • Why defense: John Carmack's 1% argument for AGI applied to defense (0:23:41–0:24:01).
  • Meta's new glasses: "I'm a huge fan… a third of the price of Apple's product… one eighth the weight" (0:30:20–0:30:33); no mention of Anduril's XR partnership with Meta (EagleEye, May 2025).
  • On founders and AI: "if you want to work on technology, work for somebody else, if you want to work on nonstop bullshit, start a company" (0:35:01); "AI doesn't actually change the bullshit part of the job very much" (0:35:06).
  • Prize: handed the XPRIZE Wildfire award (0:00:34), and announced as the funder of a $10M animal-communication XPRIZE "in a month's time" (0:03:31). See XPRIZE.

What changed between EP169 and EP295

  • Wildfire. EP169 1:26:21: "This is a matter of product execution. The tech to detect and exterminate destructive wildfires, maybe not in all of them, but… let's say 95% of them, it exists today." The test came (Alaska, June 2026): first place in the autonomous track, no team fully suppressed a fire, grand prize not awarded. EP295: "it works" 0:00:41, "possible for over a decade" 0:00:43. The claim did not move; the result did not reach the stage.
  • No other position changed. Autonomous weapons, humanoids, P(doom): the same argument, in places the same sentences.

Facts that are checkable

claim status
Co-lead of Project Meridian with Musk and Gingrich ✅ CNBC, TWZ
"$2.2 million… a million of that from Lockheed" ✅ $1.2M first place (autonomous track) + $1M Lockheed Martin detection bonus (PG&E newsroom); ⚠️ grand prize unclaimed
"you got this prize announced back in 2019" 🔴 launched 2023-04-21 (Axios)
Anduril–Meta military XR (EagleEye) ✅ announced 2025-05-29 (Mixed, eWeek); contract status after mid-2025 not checked
Anduril holds multi-billion-dollar defence contracts (Peter, EP299 2:02:40) ✅ CNBC raises the same conflict-of-interest point

Where it came up

moonshots 169 (solo guest) · Moonshots #295 (guest) · Moonshots #298 2:13:04–2:14:07 (AMA, Dave adopts his risk frame) · Moonshots #299 0:00:45, 2:01:57, 2:04:26, 2:05:09, 2:08:29 · Moonshots #303 0:50:58 (the prize, "about to"). Earlier mentions in 12 backfill transcripts (including the patents-as-instruction-manual story) and EP283/284/285/288.

Open questions

  • The $10M prize: due ~25 Oct by the host's date. Check xprize.org the week of 26 Oct.
  • What the on-screen instruction at 0:33:49 was about ("We're not going to talk about company"); unresolved, do not guess.

Related

Anduril · XPRIZE · Lockheed Martin · John Carmack · Meta · Project Meridian · Dave Blundin · Moonshots: the disclosures ledger · Moonshots #307

EP307 (Moonshots #307, aired 2026-10-31): BlueChip, and the stake that was not said on his own tape

The host says he is an investor in Anduril ("Full disclosure, I'm an investor in Anderrol." 1:53:21). It was not said on Moonshots #295, where Luckey was the guest and the host's XPRIZE paid his company. The story is the BlueChip laptop (Anduril × Qualcomm, 27 Oct): "80 years ago, America pioneered modern computing to defend our nation. Now we're bringing it home" (company video, 1:54:07). The host: "I flew with him back from uh from New York where he was winning the uh the Wildfire X-P prize" (1:55:19). Anduril won one track; the grand prize is unclaimed and New York is unconfirmed. The $10M prize he funds is still not mentioned on EP307, a week past the host's ~25 Oct date. → Moonshots: the disclosures ledger

Reading rule. Anything marked unverified or on-air is a claim made on the podcast and has not been checked against a primary source yet. It stays marked until it has. Sourced numbers link to where they came from.